Researchers at the Viterbi School of Engineering have designed brain implants that fit the exact area of a brain rather than using a standard shape. They hope their progress with the implant could one day help them build a device that restores lost memory.

According to one of the lead researchers and associate professor of neurological surgery and of biomedical engineering, Dong Song, this work is only one piece of a bigger goal: a memory prosthesis, a device meant to replace the hippocampus, the part of the brain that turns short-term memories into long-term ones.

The paper was published Aug. 12 and lays out a process for designing custom brain-recording implants for animal studies.

The research comes from the labs of Song and Ellis Meng, a professor of biomedical engineering as well as electrical and computer engineering at Viterbi.

Building such a device requires a way to record from many neurons, process the signals with a computer model and then send signals back to the brain. Song compared this process to a bridge.

“If you have a bridge, [and] it’s broken, then there’s no traffic,” Song said. “What we are trying to do is fix that bridge.”

The paper focuses on the first part of that process: creating a better connection between the device and the brain. Meng said her lab builds the devices in her lab space and in USC’s O’Brien Nanofabrication Laboratorynanofabrication core facility.

The hard part, she said, is designing the overall shape and deciding where to put the electrodes so they reach the right regions.

From the start, the teams have worked with the very scientists who will be using the devices. Song said the approach is a back-and-forth, with researchers explaining what they need and engineers explaining what is possible.

He noted that some companies are unlikely to do this because each project’s needs are specialized, and the market is small. The National Institutes of Health has funded the work for six years, and his team is now seeking to renew the grant.

In Song’s experience, different researchers want different things: Engineers often want as much signal as possible, even if it’s noisy, because artificial intelligence can help decode it. Neuroscientists often want cleaner recordings from single neurons, even if there are fewer of them. Because each group’s goals shape what it needs from a device, Song said, the lab can’t assume what a researcher wants. Balancing those needs, he said, is the biggest challenge of the work.

His answer is to talk early and often. The team asks researchers what they want to learn, explains what the devices can and can’t do, and adjusts from there.

“You really need to talk to those people,” Song said.

Beyond Song’s personal work, he says graduate students and postdoctoral researchers have been central to the project. They help build the devices and test them in animal experiments, which many engineering labs aren’t set up to do.

Two of those doctoral candidates, fifth-year Yingyi Gao and third-year Jeannie Liang, study biomedical engineering in Meng’s lab. They made the devices, from the first design to the finished implant, and then tested them.

According to Liang, the process begins on a computer. They draw each implant’s design in a computer-aided design software, then shape it before building the final device in a cleanroom.

“There are so many steps along the way that could go wrong,” Liang said. “That sometimes did go wrong.”

When it does, she said, they have started over. Still, she said, the cleanroom work drew her to the lab; she never wanted to work in software coding, she said. In the lab, she got to learn each step from design to fabrication, and even wore a bunny suit. She called it “very fun.”

Gao said she joined the lab for hands-on experimentation. She worked on neuroscience-related projects as an undergraduate and wanted to keep going in that direction. For her, the real payoff came when the implants worked in a live animal.

“I was very surprised to see it actually work in real animals,” Gao said. “The first time I saw the neural spikes, I was very surprised and ultimately impressed.”

Beyond building implants that fit the brain, the project aims to share what they’ve learned with the public and other researchers. Liang said the team wants to make its designs available to anyone online, so other labs can see the problems they ran into and use the designs in their own work. Gao hopes other labs will use their own devices with their methods and learn from them.

For now, the implants stay in animals, and a working memory prosthesis for humans is still a long way off.

“I feel like this is one step closer to building a real memory prosthesis for future human study,” Song said in a Viterbi news story about the research.